The proposed experimental and computational work is designed to explore a learning account of structural repetition in adult language use. Structural repetition is a tendency to repeat syntactic structures in successive utterances. The central hypothesis is that the tendency reflects a change in the language production system that is similar in principle to the performance effects of exposure to and use of language structures in the course of language learning. The viability of a learning account of structural repetition depends on evidence that such repetition displays accepted features of cognitive-procedural or implicit learning, and gathering such evidence is one aim of the work. A second aim concerns the more specific issue of learning to perform language itself, or learning-to-talk. A key to the viability of a learning-to-talk account of structural repetition, as opposed to a fully general learning account, is the rationality of the generalization function from one utterance to another. If the features of one sentence generalize to a subsequent sentence only when the two share certain properties, the implication is that generalizations are constrained by the shared linguistic features. A major focus of the proposed experimental work is to determine what these constraining features might be. In this way, the research begins to address questions about the interface between language knowledge and language use. The experiments employ structural priming paradigms to examine the nature of the generalizations that are made across sentences, in order to establish what utterances must have in common for the structural features of one to be reflected in another. Because the variations in sentence form which constitute structural repetition are a product of the language production system, the results of the experiments also bear on an account of normal production mechanisms. The computational modelling effort focuses on specifying the features of a representational and processing system that is capable of simulating such results, developing an explicit description of the syntactic component of language production.